CN102825603B - Network teleoperation robot system and time delay overcoming method - Google Patents

Abstract

The invention discloses a network teleoperation robot system and a time delay overcoming method. The system is formed by linking two manipulators with the same structure and symmetrical functions through a network communication link. A main manipulator is operated by an operator so that the main manipulator is acted; the main manipulator transmits an action signal to a secondary manipulator through force and angle displacement sensors, and a communication control link, so that the secondary manipulator can move in the same direction along the main manipulator. When the secondary manipulator is affected by the environment, the secondary manipulator can return the signal to the main manipulator along the original path, thus, the main manipulator can feel the same action of the environment and feed back to the operator. By adopting a modeling method based on an event and a fuzzy predictive control algorithm, the system has good robustness on the model mismatch; therefore, the system can overcome the influence caused by the network time delay.

Description

Internet-based telerobotics system and time delay conquering method

Technical field

The present invention relates to a kind of control system and method for robot, particularly relate to a kind of control system and time delay conquering method of network teleoperation robot, belong to robotics.

Background technology

Since generation in 20th century 90, computer networking technology obtains develop rapidly, and computer network, mechano-electronic are connected robot with the technology of sensor aspect with network, can realize network teleoperation robot.Internet-based telerobotics can utilize network as medium, connect equipment such as being positioned at the operator at network two ends and robot, operator utilizes the status information such as sound, image, position, power distally fed back to, in real time interactive controlling is carried out to REMOTE MACHINE people, complete the operation behaviors such as production, experiment, exploration.

But, along with teleoperation robot is widely applied to the every field of space, deep-sea, industrial production and people's lives, while the security and operating efficiency raising of operator, its serious deficiency also comes out: the intrinsic delay character of remote control system affects the normal work of system, during comparatively large the or change of time delay, by greatly reducing the performance of system, even cause instability.The network transfer delay especially impact that brings to teleoperation robot of long time delay and time-varying delay, is mainly reflected in the stability of reduction system, the transparency.

Through the development of decades, scholar's brainstrust both domestic and external has proposed many methods solving delay problem, and the research method at present about Teleoperation Systems mainly contains three kinds, i.e. PREDICTIVE CONTROL, remote layout and bilateral control.

PREDICTIVE CONTROL is as the effective ways solving remote control system delay problem, and its main thought is the dummy model by setting up at main robot control station from machine human and environment, and predicts from manipulator is stressed with dummy model.When the model of remote Slave device human and environment is identical with the model of dummy model, operator is just equal to and the contacting of true environment the contact of dummy model.But PREDICTIVE CONTROL also has the weak point of self, its a kind of accuracy control method based on model, needs the accurate model known from manipulator and environment, and control algolithm relative complex.In order to solve the problem of predictive control algorithm complexity, PREDICTIVE CONTROL is combined with fuzzy reasoning, can by pattern die gelatinization, make control method simpler, more meet the control thought of people, combine again and do not rely on the modeling method of time based on event, the impact of time delay can be overcome further, improve control effects.

Summary of the invention

The object of the present invention is to provide a kind of Internet-based telerobotics system and time delay conquering method, optimized network remote control system, and solve the existing delay problem of Internet-based telerobotics.

Object of the present invention is achieved by the following technical programs:

A kind of Internet-based telerobotics system, identical by two structures, the manipulator of function symmetry is linked together by network communication link.Comprise main frame mechanical arm 1, from mechanical arm 2, master computer 3, from computer 4, first data collecting card 5, second data collecting card 6, first single-chip microcomputer 7, second singlechip 8.Described master computer 3 gathers position and the force signal of main frame mechanical arm 1 by the first data collecting card 5, describedly gather position from mechanical arm 2 and force signal from computer 4 by the second data collecting card 6, described master computer 3 and by TCP/IP network communication link, the position of slave mechanical arm and force signal being transmitted mutually from computer 4, control signal is passed to the first single-chip microcomputer 7 by serial communication by described master computer 3, first single-chip microcomputer 7 exports pwm pulse signal main control system mechanical arm 1 and works, describedly from computer 4, control signal is passed to second singlechip 8 by serial communication, second singlechip 8 exports pwm pulse signal and controls to work from mechanical arm 2.

A time delay conquering method for Internet-based telerobotics system, comprises the following steps:

1. set up for network delay based on event model

Replace the reference variable-time T in existing system with event variable s, in event-based control, choosing s is the distance that manipulator is passed by, and sets whenever s increases progressively a segment distance, produces an event, and current status information is sent to path management device;

The production method of locomotion state: obtain feedback event s through event generator, in order to make that reference event s is expected in reference event s tracking _d, adopt following algorithm to produce the event input of expectation:

$s_d=s-k(\frac{{\mathrm{ds}}_d}{\mathrm{dt}}-\frac{\mathrm{ds}}{\mathrm{dt}})$

Definition e=s _d-s, known s converges on s _d;

If the universal model of robot is as follows:

$\left\{\begin{array}{cc}\frac{\mathrm{dx}}{\mathrm{dt}}=f\left(x\right)+g\left(x\right)u& x\∈R^n,u\∈R^m\\ y=h\left(x\right)& y\∈R^m\end{array}\right.$

New reference variable s=S (x (t)) selected by order, by the known robot system based on event model can be expressed as follows:

$\left\{\begin{array}{cc}\frac{\mathrm{dx}}{\mathrm{dt}}=\frac{f\left(x\left(s\right)\right)}{v\left(s\right)}+\frac{g\left(x\left(s\right)\right)}{v\left(s\right)}u& x\∈R^n,u\∈R^m\\ y=h\left(x\left(s\right)\right)& y\∈R^m\end{array}\right.$

Wherein, $v\left(s\right)=\frac{\mathrm{dS}\left(x\left(t\right)\right)}{\mathrm{dt}}$

By the model parameter of the known system of the model based on event set up;

2. fuzzy prediction control algorithm

In conjunction with the set up design carrying out Fuzzy Generalized predictive controller based on event model;

Model prediction and rolling optimization based on generalized predictive control:

If the forecast model of Internet-based telerobotics system is by controlled autoregressive integrated moving average model, namely CARIMA equation describes:

A(z ^-1)y(k)=B(z ^-1)u _GPC(k-1)+C(z ^-1)ξ(k)/Δ

Wherein A (z ^-1), B (z ^-1) and C (z ^-1) be the z on n, m and n rank respectively ^-1multinomial, C (z can be made ^-1)=1, y (k) is system output, u _gPC(k-1) represent controlled quentity controlled variable, ξ (k) represents that average is the white noise sequence of zero, z ^-1for backward shift operator, Δ=1-z ^-1for difference operator, if time lag is greater than zero, then B (z ^-1) multinomial beginning one or several systems equal zero.Because CARIMA model can be included in control law integral action naturally, can be eliminated the steady-state error of system;

Consider the optimality criterion in k moment, adopt following formula to represent:

$J=\underset{j=1}{\overset{N_1}{\mathrm{\Σ}}}{[y(k+j)-y_r(k+j)]}^2+\underset{j=1}{\overset{N_{\mathrm{\μ}}}{\mathrm{\Σ}}}\mathrm{\λ}{\left[{\mathrm{\Δu}}_{\mathrm{GPC}}(k+j-1)\right]}^2$

In formula, y _rfor known reference sequences, λ be greater than zero control weight coefficient, N ₁maximum predicted time domain, N _μrepresent and control time domain (N _μ<N ₁), namely at N _μafter step, controlled quentity controlled variable will no longer change;

In order to predict the output that advanced j walks, introduce Diophantine equation

1＝E _j(z ^-1)A(z ^-1)Δ+z ^-jF _j(z ^-1)

E _j(z ^-1)B _j(z ^-1)=G _j(z ^-1)+z ^-jH _j(z ^-1)

Wherein E _j(z ^-1), F _j(z ^-1), G _j(z ^-1) and H _j(z ^-1) be the z on j-1, n, j-1 and m-1 rank respectively ^-1multinomial;

As shown from the above formula, as long as given prediction time domain N ₁, control time domain N _μwith weighting constant λ, controlled quentity controlled variable Δ u just can be obtained _gPC(k), its vector form is as follows:

u _GPC(k)=u(k-1)+g ^T(y _r-Fy)

Δu _GPC(k)=g ^T(y _r-Fy)

In formula, g ^tfor (G ^tg+ λ I) ^-1g ^tthe first row;

Specific algorithm step is as follows:

The first step: by the known plant model parameter A of the model (z based on event ^-1) and B (z ^-1);

Second step: given prediction time domain N ₁, control time domain N _μwith weighting constant λ;

3rd step: by Diophantine Equation Solution multinomial E _j, F _j, G _jand H _j;

4th step: compute matrix G and (G ^tg+ λ I) ^-1;

5th step: solve controlled quentity controlled variable u _gPC(k) and Δ u _gPC(k).

Object of the present invention can also be realized further by following technical measures:

The time delay conquering method of aforementioned network Teleoperation Systems, also comprises the feedback correction method based on fuzzy reasoning;

Control signal Δ u (k) is made up of two parts below:

Δu(k)=Δu _GPC(k)+Δu _F(k)

Wherein u _fwhat k () was fuzzy reasoning obtains compensation of error controlled quentity controlled variable;

If the feedback deviation that e (k) and ec (k) is the system k moment and change of error value, walk time delay, then u because system exists τ _fk () is determined by e (k-τ) and ec (k-τ);

That is: e (k-τ)=y _r(k-τ)-y _m(k-τ)

ec(k-τ)=e(k-τ)-e(k-τ)

Wherein, y _r(k-τ) is object output feedack value, y _m(k-τ) is object model output valve, then u _fcan be judged by e (k-τ) and ec (k-τ);

The design of fuzzy compensation controller will adopt following control law: the domain of e and ec is divided into respectively 5 fuzzy set { NB, NS, ZE, PS, PB} and 3 fuzzy set { N, Z, P}, compensate to obtain predicated error more accurately, setting deviation e adopts Gauss's membership function, deviation variation rate ec adopts Triangleshape grade of membership function, and ambiguity in definition rule is as follows:

R _i：If?e?is?A _m?and?ec?is?B _n?then?u _F?is?C _k

Wherein, A _m∈ { NB, NS, ZE, PS, PB}, B _n∈ N, Z, P}, i=1,2 ..., 15, u _ffor compensatory control input;

Control law is as follows:

When carrying out anti fuzzy method, in order to obtain more accurate controlled quentity controlled variable, area gravity model appoach is adopted to try to achieve u _fclear value, namely calculate with following inference method:

$u_F=\underset{i=1}{\overset{15}{\mathrm{\&Sigma;}}}({\stackrel{\&OverBar;}{w}}_i\&CenterDot;C_i)/\underset{i=1}{\overset{15}{\mathrm{\&Sigma;}}}{\stackrel{\&OverBar;}{w}}_i$

$w_i=A_m\left(e(k-\mathrm{\&tau;})\right)\&CenterDot;B_n\left(\mathrm{ec}(k-\mathrm{\&tau;})\right)/\underset{i=1}{\overset{15}{\mathrm{\&Sigma;}}}(A_m\left(e(k-\mathrm{\&tau;})\right)\&CenterDot;B_n\left(\mathrm{ec}(k-\mathrm{\&tau;})\right))$

Consider u _fchanging excessively may affect control effects, then error compensation control amount Δ u _fk () is retrained by following formula:

Δu _F(k)=λ(u _F(k)-u _F(k-1))

Wherein λ ∈ [0,1], is constant.

Compared with prior art, the invention has the beneficial effects as follows:

1. the present invention is directed to the delay problem that network tele-operation system exists, have employed the model based on event, evade the impact of delay problem in modeling, make system become closed loop, real-time control system, can ensure the stability of system.

2., setting up based on the basis of event model, PREDICTIVE CONTROL being combined with fuzzy reasoning, has good robustness to model mismatch, also can overcome the impact of time delay better.

Accompanying drawing explanation

Fig. 1 is Internet-based telerobotics system construction drawing of the present invention;

Fig. 2 is that Internet-based telerobotics principal and subordinate holds control structure figure;

Fig. 3 is LabVIEW main side software flow pattern;

Fig. 4 is that LabVIEW is from end software flow pattern;

Fig. 5 is LabVIEW front panel design block diagram;

Fig. 6 is the model framework chart based on event;

Fig. 7 is the structured flowchart of Fuzzy Predictive Controller.

Detailed description of the invention

Below in conjunction with the drawings and specific embodiments, the invention will be further described.

As shown in Figure 1, Internet-based telerobotics system, identical by two structures, the manipulator of function symmetry is linked together by network communication link.

As shown in Figure 2, Internet-based telerobotics system, comprise main frame mechanical arm 1, from mechanical arm 2, master computer 3, from computer 4, first data collecting card 5, second data collecting card 6, first single-chip microcomputer 7, second singlechip 8.Described master computer 3 gathers position and the force signal of main frame mechanical arm 1 by the first data collecting card 5, describedly gather position from mechanical arm 2 and force signal from computer 4 by the second data collecting card 6, described master computer 3 and by TCP/IP network communication link, the position of slave mechanical arm and force signal being transmitted mutually from computer 4, control signal is passed to the first single-chip microcomputer 7 by serial communication by described master computer 3, first single-chip microcomputer 7 exports pwm pulse signal main control system mechanical arm 1 and works, describedly from computer 4, control signal is passed to second singlechip 8 by serial communication, second singlechip 8 exports pwm pulse signal and controls to work from mechanical arm 2.

Operator handles the action of main frame mechanical arm, consequent displacement signal is gathered by data collecting card by angular position pick up, and by TCP/IP network communication link, position signalling is delivered to from hand computer, from computer by certain algorithm, control signal is passed to single-chip microcomputer by serial communication, single-chip microcomputer sends the rotation of pwm pulse signal control from mechanical arm motor, enables to follow the motion of main frame mechanical arm from mechanical arm.When running into the effect of environment from mechanical arm, force snesor will experience force signal, and gathered by data collecting card, master computer is sent to by TCP/IP network communication link, master computer is by certain algorithm, and control signal is passed to single-chip microcomputer by serial communication, and single-chip microcomputer sends the rotation of pwm pulse signal main control system mechanical arm motor again, make operator experience from the power suffered by mechanical arm, action is below adjusted.

Software platform on computer by LabVIEW software building, will realize principal and subordinate's hand position tracking test, and the various waveform of display experiment in real time on computers and data.The LabVIEW software platform of Internet-based telerobotics mainly comprises the design of front panel and flow chart.Flow chart will comprise the design of four parts below.

(1) data acquisition: in Internet-based telerobotics system, needs the analog data of position by data collecting card harvester mechanical arm and power.Therefore, in LabVIEW platform, set up a simple VI, the analog quantity realizing data collecting card reads.

(2) serial communication: in Internet-based telerobotics system, single-chip microcomputer needs the instruction sent by serial ports receiving computer, thus produces corresponding pwm pulse signal control motor, drives slave mechanical arm to realize position and follows the tracks of and force feedback.

(3) network communication: in this experimental system, master-slave computer adopts ICP/IP protocol to be transmitted mutually the data such as the position power of slave mechanical arm by network interface.Control to follow the tracks of main hands movement from hand from computer according to slave mechanical arm position difference; Master computer feeds back from the power effect suffered by hand according to the difference of slave mechanical arm power.

(4) algorithm design: Internet-based telerobotics claimed accuracy is high, accurate positioning, stable operation, PID as classical control method controls to be difficult to reach actual requirement under the disturbing factors such as network delay, so the present invention proposes the Fuzzy Predictive Control scheme of Internet-based telerobotics system LabVIEW experiment porch.Under LabVIEW platform, do not provide the kit of fuzzy control and PREDICTIVE CONTROL, but LabVIEW has the function of script node, can external file be performed by script node user, as the m file of MATLAB.The function panel of LabVIEW software block diagram can import the control algolithm file that MATLAB script node is finished writing.LabVIEW software flow pattern as shown in Figure 3,4.

LabVIEW programming for Internet-based telerobotics is divided into main frame mechanical arm end and from mechanical arm end two parts.Configure relevant parameters from mechanical arm end, the main frame mechanical arm port that end is being specified is intercepted, and when sending TCP connection request from end, main side connects, and now system starts to carry out network communication.Positional information sends to from end by data acquisition by main side, reads data and show preservation from end, simultaneously by certain algorithm, data is sent to single-chip microcomputer by serial communication, makes the motion following the tracks of main frame mechanical arm from mechanical arm.When running into environmental activity, force information will be sent to main side by data acquisition from end, main side is read data and is shown preservation, and data are sent to single-chip microcomputer by serial communication, and main side is experienced from holding the power received.In the process, main side can ask to terminate to connect by sending local stopping.Receive " stopping " mark of main side from end after, also can close local TCP and connect.

As shown in Figure 5, three parts will mainly be comprised in LabVIEW front panel.Part I is address, the port of network connection and connects stop button etc.; Part II is disc type principal and subordinate hand angle value and bar formula principal and subordinate hand-power value; Part III is the principal and subordinate's hand position curve or principal and subordinate's hand-power value curve that export, can facilitate the real-time tracking feedback of observing system.

Internet-based telerobotics time delay conquering method is specific as follows:

1. setting up based on event model for network delay

As shown in Figure 6, in order to solve network delay problem better, present invention employs the modeling pattern based on event.

It is relevant and with time irrelevant variable as locomotion state, planning and the design of system are all carry out based on this new event variable s that to be exactly selection one to system export model essence based on event.S replaces the reference variable-time T in system in the past, thus evade the impact of delay problem, and make system become closed loop, real-time control system, can ensure the stability of system.When system cloud gray model runs into uncertain, sudden behaviour part in load environment or in running, based on event model also can keeping system coordination and possess process these sudden situations ability, further ensure the stability of system.

From Lyapunov stability criteria, if former robot system is asymptotically stability when time t is reference, so from inverse theorem, we can find a Lyapunov function L (X (t)) meeting:

1.L (X (t)) positive definite

2. negative definite.

If the motion of system is using event s as reference variable, and s=П (y) is the nondecreasing function of time variable t, so L (X (s)) still positive definite.

In addition, $\frac{\mathrm{dL}\left(X\left(t\right)\right)}{\mathrm{dt}}=\frac{\mathrm{dL}\left(X\left(s\right)\right)}{\mathrm{dt}}=\frac{\mathrm{dL}\left(X\left(s\right)\right)}{\mathrm{ds}}\frac{\mathrm{ds}}{\mathrm{dt}}$

Subtract or monotonic increasing function as long as selected event s is the non-of time t, so

Therefore it is negative definite.So this system is about event reference variable s asymptotically stability.

As can be seen here, the reliability of event-based control system is that event model does not directly depend on the time, and thus time delay can not have an impact to the stable of control system.That is, as long as find suitable event s to be reference variable for system, delay problem will reduce greatly on the impact of control system.

In event-based control, choosing s is the distance that master manipulator is passed by, owing to expecting that s is the function increased progressively in time, therefore the event selected meets the demands. set whenever s increases progressively the first distance, during as 0.02m, an event will be produced, and current status information is sent to path management device.

The production method of locomotion state: obtain feedback event s through event generator, in order to make that reference event s is expected in reference event s tracking _d, adopt following algorithm to produce the event input of expectation:

$s_d=s-k(\frac{{\mathrm{ds}}_d}{\mathrm{dt}}-\frac{\mathrm{ds}}{\mathrm{dt}})$

Definition e=s _d-s, known s converges on s _d.

The universal model of robot is as follows:

$\left\{\begin{array}{cc}\frac{\mathrm{dx}}{\mathrm{dt}}=f\left(x\right)+g\left(x\right)u& x\&Element;R^n,u\&Element;R^m\\ y=h\left(x\right)& y\&Element;R^m\end{array}\right.$

New reference variable s=S (x (t)) selected by order, by the known robot system based on event model can be expressed as follows:

$\left\{\begin{array}{cc}\frac{\mathrm{dx}}{\mathrm{dt}}=\frac{f\left(x\left(s\right)\right)}{v\left(s\right)}+\frac{g\left(x\left(s\right)\right)}{v\left(s\right)}u& x\&Element;R^n,u\&Element;R^m\\ y=h\left(x\left(s\right)\right)& y\&Element;R^m\end{array}\right.$

Wherein, $v\left(s\right)=\frac{\mathrm{dS}\left(x\left(t\right)\right)}{\mathrm{dt}}$

By the model parameter of the known system of the model based on event set up.

2. fuzzy prediction control algorithm

As shown in Figure 7, the present invention has considered the prediction of output of PREDICTIVE CONTROL and the feature of rolling optimization, to the error that model mismatch causes, directly adopts fuzzy reasoning, carries out feedback compensation.Like this, combine the advantage of PREDICTIVE CONTROL and fuzzy reasoning, the puzzlement of delay problem to Internet-based telerobotics can be solved further.

In conjunction with set up based on event model, carry out the design of Fuzzy Generalized predictive controller below.

1) based on model prediction and the rolling optimization of generalized predictive control

If the forecast model of Internet-based telerobotics system is by controlled autoregressive integrated moving average model, namely CARIMA equation describes:

A(z ^-1)y(k)=B(z ^-1)u _GPC(k-1)+C(z ^-1)ξ(k)/Δ

Wherein A (z ^-1), B (z ^-1) and C (z ^-1) be the z on n, m and n rank respectively ^-1multinomial, C (z can be made ^-1)=1, y (k) is system output, u _gPC(k-1) represent controlled quentity controlled variable, ξ (k) represents that average is the white noise sequence of zero, z ^-1for backward shift operator, Δ=1-z ^-1for difference operator, if time lag is greater than zero, then B (z ^-1) multinomial beginning one or several systems equal zero.Because CARIMA model can be included in control law integral action naturally, can be eliminated the steady-state error of system.

Consider the optimality criterion in k moment, adopt following formula to represent:

$J=\underset{j=1}{\overset{N_1}{\mathrm{\&Sigma;}}}{[y(k+j)-y_r(k+j)]}^2+\underset{j=1}{\overset{N_{\mathrm{\&mu;}}}{\mathrm{\&Sigma;}}}\mathrm{\&lambda;}{\left[{\mathrm{\&Delta;u}}_{\mathrm{GPC}}(k+j-1)\right]}^2$

In formula, y _rfor known reference sequences, λ be greater than zero control weight coefficient, N ₁maximum predicted time domain, N _μrepresent and control time domain (N _μ<N ₁), namely at N _μafter step, controlled quentity controlled variable will no longer change.

In order to predict the output that advanced j walks, introduce Diophantine equation

1＝E _j(z ^-1)A(z ^-1)Δ+z ^-jF _j(z ^-1)

E _j(z ^-1)B _j(z ^-1)=G _j(z ^-1)+z ^-jH _j(z ^-1)

Wherein E _j(z ^-1), F _j(z ^-1), G _j(z ^-1) and H _j(z ^-1) be the z on j-1, n, j-1 and m-1 rank respectively ^-1multinomial.

As shown from the above formula, as long as given prediction time domain N ₁, control time domain N _μwith weighting constant λ, controlled quentity controlled variable Δ u just can be obtained _gPC(k).Its vector form is as follows:

u _GPC(k)=u(k ^-1)+g ^T(y _r-Fy)

Δu _GPC(k)=g ^T(y _r-Fy)

In formula, g ^tfor (G ^tg+ λ I) ^-1g ^tthe first row.

Specific algorithm step is as follows:

The first step: by the known plant model parameter A of the model (z based on event ^-1) and B (z ^-1).

Second step: given prediction time domain N ₁, control time domain N _μwith weighting constant λ

3rd step: by Diophantine Equation Solution multinomial E _j, F _j, G _jand H _j.

4th step: compute matrix G and (G ^tg+ λ I) ^-1

5th step: solve controlled quentity controlled variable u _gPC(k) and Δ u _gPC(k).

2) based on the feedback compensation of fuzzy reasoning

In general generalized predictive control, forecast model and the rolling optimization thought of PREDICTIVE CONTROL are obtained for embodiment, and the embodiment of feedback compensation link is less.Due to environment, noise, network delay and disturbing factor, the Mathematical Modeling of actual control system and forecast model is made to have larger error, in order to address this problem, the present invention utilizes fuzzy compensation to revise the output of forecast model, and it can not only overcome the impact of model mismatch further, but also improve the accuracy of prediction, namely control signal Δ u (k) is made up of two parts below:

Δu(k)=Δu _GPC(k)+Δu _F(k)

Wherein u _fwhat k () was fuzzy reasoning obtains compensation of error controlled quentity controlled variable.

If the feedback deviation that e (k) and ec (k) is the system k moment and change of error value, walk time delay, then u because system exists τ _fk () is determined by e (k-τ) and ec (k-τ).

That is: e (k-τ)=y _r(k-τ)-y _m(k-τ)

ec(k-τ)=e(k-τ)-e(k-τ-1)

Wherein, y _r(k-τ) is object output feedack value, y _m(k-τ) is object model output valve, then u _fcan be judged by e (k-τ) and ec (k-τ).

The design of fuzzy compensation controller will adopt following control law: the domain of e and ec is divided into respectively 5 fuzzy set { NB, NS, ZE, PS, PB} and 3 fuzzy set { N, Z, P}, compensate to obtain predicated error more accurately, setting deviation e adopts Gauss's membership function, deviation variation rate ec adopts Triangleshape grade of membership function, and ambiguity in definition rule is as follows:

R _i：If?e?is?A _m?and?ec?is?B _n?then?u _F?is?C _k

Wherein, A _m∈ { NB, NS, ZE, PS, PB}, B _n∈ N, Z, P}, i=1,2 ..., 15, u _ffor compensatory control input.

Control law is as follows:

When carrying out anti fuzzy method, in order to obtain more accurate controlled quentity controlled variable, employing area gravity model appoach is tried to achieve u _fclear value, namely calculate with following inference method:

$u_F=\underset{i=1}{\overset{15}{\mathrm{\&Sigma;}}}({\stackrel{\&OverBar;}{w}}_i\&CenterDot;C_i)/\underset{i=1}{\overset{15}{\mathrm{\&Sigma;}}}{\stackrel{\&OverBar;}{w}}_i$

Wherein, $w_i=A_m\left(e(k-\mathrm{\&tau;})\right)\&CenterDot;B_n\left(\mathrm{ec}(k-\mathrm{\&tau;})\right)/\underset{i=1}{\overset{15}{\mathrm{\&Sigma;}}}(A_m\left(e(k-\mathrm{\&tau;})\right)\&CenterDot;B_n\left(\mathrm{ec}(k-\mathrm{\&tau;})\right))$

In addition, consider and prevent because of u _fchanging excessively affects control effects, error compensation control amount Δ u _fk () is retrained by following formula:

Δu _F(k)=λ(u _F(k)-u _F(k-1))

Wherein λ ∈ [0,1], is constant.

In addition to the implementation, the present invention can also have other embodiments, and all employings are equal to the technical scheme of replacement or equivalent transformation formation, all drop in the protection domain of application claims.

Claims (1)

1. an Internet-based telerobotics Time Delay of Systems conquering method, Internet-based telerobotics system is identical by two structures, and the manipulator of function symmetry is linked together by network communication link, comprise main frame mechanical arm (1), from mechanical arm (2), master computer (3), from computer (4), the first data collecting card (5), the second data collecting card (6), the first single-chip microcomputer (7), second singlechip (8), described master computer (3) gathers position and the force signal of main frame mechanical arm (1) by the first data collecting card (5), describedly gather from the position of mechanical arm (2) and force signal from computer (4) by the second data collecting card (6), described master computer (3) and by TCP/IP network communication link, the position of slave mechanical arm and force signal being transmitted mutually from computer (4), control signal is passed to the first single-chip microcomputer (7) by serial communication by described master computer (3), first single-chip microcomputer (7) exports pwm pulse signal main control system mechanical arm (1) work, describedly from computer (4), control signal is passed to second singlechip (8) by serial communication, second singlechip (8) exports pwm pulse signal and controls to work from mechanical arm (2),

The time delay conquering method of Internet-based telerobotics system, comprises the following step:

1) set up for network delay based on event model

Replace the reference variable-time T in existing system with event variable s, in event-based control, choosing s is the distance that manipulator is passed by, and sets whenever s increases progressively a segment distance, produces an event, and current status information is sent to path management device;

The production method of locomotion state: obtain feedback event s through event generator, in order to make that reference event s is expected in reference event s tracking _d, adopt following algorithm to produce the event input of expectation:

$s_d=s-k(\frac{{\mathrm{ds}}_d}{\mathrm{dt}}-\frac{\mathrm{ds}}{\mathrm{dt}})$

Definition e=s _d-s, known s converges on s _d;

If the universal model of robot is as follows:

$\left\{\begin{array}{cc}\frac{\mathrm{dx}}{\mathrm{dt}}=f\left(x\right)+g\left(x\right)u& x\&Element;R^n,u\&Element;R^m\\ y=h\left(x\right)& y\&Element;R^m\end{array}\right.$

New reference variable s=S (x (t)) selected by order, by the known robot system based on event model can be expressed as follows:

$\left\{\begin{array}{cc}\frac{\mathrm{dx}}{\mathrm{ds}}=\frac{f\left(x\left(s\right)\right)}{v\left(s\right)}+\frac{g\left(x\left(s\right)\right)}{v\left(s\right)}u& x\&Element;R^n,u\&Element;R^m\\ y=h\left(x\left(s\right)\right)& y\&Element;R^m\end{array}\right.$

Wherein, $v\left(s\right)=\frac{\mathrm{dS}\left(x\left(t\right)\right)}{\mathrm{dt}}$

By the model parameter of the known system of the model based on event set up;

2) fuzzy prediction control algorithm

In conjunction with the set up design carrying out Fuzzy Generalized predictive controller based on event model;

Model prediction and rolling optimization based on generalized predictive control:

If the forecast model of Internet-based telerobotics system is by controlled autoregressive integrated moving average model, namely CARIMA equation describes:

A(z ^-1)y(k)＝B(z ^-1)u _GPC(k-1)+C(z ^-1)ξ(k)/Δ

Wherein A (z ^-1), B (z ^-1) and C (z ^-1) be the z on n, m and n rank respectively ^-1multinomial, C (z can be made ^-1)=1, y (k) is system output, u _gPC(k-1) represent controlled quentity controlled variable, ξ (k) represents that average is the white noise sequence of zero, z ^-1for backward shift operator, Δ=1-z ^-1for difference operator, if time lag is greater than zero, then B (z ^-1) multinomial beginning one or a few term coefficient equal zero, because CARIMA model can be included in control law integral action naturally, can be eliminated the steady-state error of system;

Consider the optimality criterion in k moment, adopt following formula to represent:

$J=\underset{j=1}{\overset{N_1}{\mathrm{\&Sigma;}}}{[y(k+j)-y_r(k+j)]}^2+\underset{j=1}{\overset{N_{\mathrm{\&mu;}}}{\mathrm{\&Sigma;}}}\mathrm{\&lambda;}{\left[\mathrm{\&Delta;}{u}_{\mathrm{GPC}}(k+j-1)\right]}^2$

In formula, y _rfor known reference sequences, λ be greater than zero control weight coefficient, N ₁maximum predicted time domain, N _μrepresent and control time domain (N _μ< N ₁), namely at N _μafter step, controlled quentity controlled variable will no longer change;

In order to predict the output that advanced j walks, introduce Diophantine equation

1＝E _j(z ^-1)A(z ^-1)Δ+z ^-jF _j(z ^-1)

E _j(z ^-1)B _j(z ^-1)＝G _j(z ^-1)+z ^-jH _j(z ^-1)

Wherein E _j(z ^-1), F _j(z ^-1), G _j(z ^-1) and H _j(z ^-1) be the z on j-1, n, j-1 and m-1 rank respectively ^-1multinomial;

As shown from the above formula, as long as given prediction time domain N ₁, control time domain N _μwith weighting constant λ, controlled quentity controlled variable Δ u just can be obtained _gPC(k), its vector form is as follows:

u _GPC(k)＝u(k-1)+g ^T(y _r-Fy)

Δu _GPC(k)＝g ^T(y _r-Fy)

In formula, g ^tfor (G ^tg+ λ I) ^-1g ^tthe first row;

Specific algorithm step is as follows:

The first step: by the known plant model parameter A of the model (z based on event ^-1) and B (z ^-1);

Second step: given prediction time domain N ₁, control time domain N _μwith weighting constant λ;

3rd step: by Diophantine Equation Solution multinomial E _j, F _j, G _jand H _j;

4th step: compute matrix G and (G ^tg+ λ I) ^-1;

5th step: solve controlled quentity controlled variable u _gPC(k) and Δ u _gPC(k);

3) the time delay conquering method of described Internet-based telerobotics system also comprises the feedback correction method based on fuzzy reasoning;

Control signal Δ u (k) is made up of two parts below:

Δu(k)＝Δu _GPC(k)+Δu _F(k)

Wherein u _fwhat k () was fuzzy reasoning obtains compensation of error controlled quentity controlled variable;

If the feedback deviation that e (k) and ec (k) is the system k moment and change of error value, walk time delay, then u because system exists τ _fk () is determined by e (k-τ) and ec (k-τ);

That is: e (k-τ)=y _r(k-τ)-y _m(k-τ)

ec(k-τ)＝e(k-τ)-e(k-τ)

Wherein, y _r(k-τ) is object output feedack value, y _m(k-τ) is object model output valve, then u _fcan be judged by e (k-τ) and ec (k-τ);

It is characterized in that, the design of fuzzy compensation controller adopts following control law: the domain of e and ec is divided into respectively 5 fuzzy sets { NB, NS, ZE, PS, PB} and 3 fuzzy set { N, Z, P}, compensate to obtain predicated error more accurately, setting deviation e adopts Gauss's membership function, and deviation variation rate ec adopts Triangleshape grade of membership function, and ambiguity in definition rule is as follows:

R _i：If?e?is?A _m?and?ec?is?B _n?then?u _F?is?C _k

Wherein, A _m∈ { NB, NS, ZE, PS, PB}, B _n∈ N, Z, P}, i=1,2 ..., 15, u _ffor compensatory control input;

Control law is as follows:

When carrying out anti fuzzy method, in order to obtain more accurate controlled quentity controlled variable, area gravity model appoach is adopted to try to achieve u _fclear value, namely calculate with following inference method:

$u_F=\underset{i=1}{\overset{15}{\mathrm{\&Sigma;}}}({\stackrel{\&OverBar;}{w}}_i\&CenterDot;C_i)/\underset{i=1}{\overset{15}{\mathrm{\&Sigma;}}}{\stackrel{\&OverBar;}{w}}_i$

$w_i=A_m\left(e(k-\mathrm{\&tau;})\right)\&CenterDot;B_n\left(\mathrm{ec}(k-\mathrm{\&tau;})\right)/\underset{i=1}{\overset{15}{\mathrm{\&Sigma;}}}(A_m\left(e(k-\mathrm{\&tau;})\right)\&CenterDot;B_n\left(\mathrm{ec}(k-\mathrm{\&tau;})\right))$

Consider u _fchanging excessively may affect control effects, then error compensation control amount Δ u _fk () is retrained by following formula:

Δu _F(k)＝λ(u _F(k)-u _F(k-1))

Wherein λ ∈ [0,1], is constant.